linux Disk Sequential Writing and Random Writing Methods
- 2021-07-01 08:42:42
- OfStack
1. Preface
Random writing will cause the magnetic head to change lanes constantly, resulting in a great reduction in efficiency; Sequential write heads hardly need to change lanes, or the time for changing lanes is very short
This article discusses the specific differences between the two in 1 and the corresponding kernel calls
2. Environmental preparedness
| 组件 | 版本 |
|---|---|
| OS | Ubuntu 16.04.4 LTS |
| fio | 2.2.10 |
3. Introduction to fio
Through the fio test, we can reflect the status in reading and writing. We need to focus on several key indicators in the output report of fio:
slat: Refers to the time from the submission of I/O to the actual execution of I/O (ES22latency)
clat: Refers to the time from I/O submission to I/O completion (Completion latency)
lat: Refers to the total time from fio creating I/O to I/O completing
bw: Throughput
iops: Number of I/O per second
4. Write tests synchronously
(1) Synchronous random writing
fio is the main test tool. In order to see the system call, using strace tool, the command looks like this:
Let's test a random write first
strace -f -tt -o /tmp/randwrite.log -D fio -name=randwrite -rw=randwrite \
-direct=1 -bs=4k -size=1G -numjobs=1 -group_reporting -filename=/tmp/test.dbExtract key information
root@wilson-ubuntu:~# strace -f -tt -o /tmp/randwrite.log -D fio -name=randwrite -rw=randwrite \
> -direct=1 -bs=4k -size=1G -numjobs=1 -group_reporting -filename=/tmp/test.db
randwrite: (g=0): rw=randwrite, bs=4K-4K/4K-4K/4K-4K, ioengine=sync, iodepth=1
fio-2.2.10
Starting 1 process
...
randwrite: (groupid=0, jobs=1): err= 0: pid=26882: Wed Aug 14 10:39:02 2019
write: io=1024.0MB, bw=52526KB/s, iops=13131, runt= 19963msec
clat (usec): min=42, max=18620, avg=56.15, stdev=164.79
lat (usec): min=42, max=18620, avg=56.39, stdev=164.79
...
bw (KB /s): min=50648, max=55208, per=99.96%, avg=52506.03, stdev=1055.83
...
Run status group 0 (all jobs):
WRITE: io=1024.0MB, aggrb=52525KB/s, minb=52525KB/s, maxb=52525KB/s, mint=19963msec, maxt=19963msec
Disk stats (read/write):
...
sda: ios=0/262177, merge=0/25, ticks=0/7500, in_queue=7476, util=36.05%
Lists the information we need to focus on:
(1) clat, the average duration is about 56ms
(2) lat, the average duration is about 56ms
(3) bw, throughput, about 52M
Let's look at the kernel call information again:
root@wilson-ubuntu:~# more /tmp/randwrite.log
...
26882 10:38:41.919904 lseek(3, 665198592, SEEK_SET) = 665198592
26882 10:38:41.919920 write(3, "\220\240@\6\371\341\277>\0\200\36\31\0\0\0\0\202\2\7\320\343\6H\26P\340\277\370\330\30e\30"..., 4096) = 4096
26882 10:38:41.919969 lseek(3, 4313088, SEEK_SET) = 4313088
26882 10:38:41.919985 write(3, "\220\240@\6\371\341\277>\0\200\36\31\0\0\0\0\202\2\7\320\343\6H\26P\340\277\370\330\30e\30"..., 4096) = 4096
26882 10:38:41.920032 lseek(3, 455880704, SEEK_SET) = 455880704
26882 10:38:41.920048 write(3, "\220\240@\6\371\341\277>\0\200\36\31\0\0\0\0\202\2\7\320\343\6H\26P\340\277\370\330\30e\30"..., 4096) = 4096
26882 10:38:41.920096 lseek(3, 338862080, SEEK_SET) = 338862080
26882 10:38:41.920112 write(3, "\220\240@\6\371\341\277>\0\2402\24\0\0\0\0\202\2\7\320\343\6H\26P\340\277\370\330\30e\30"..., 4096) = 4096
26882 10:38:41.920161 lseek(3, 739086336, SEEK_SET) = 739086336
26882 10:38:41.920177 write(3, "\220\240@\6\371\341\277>\0\2402\24\0\0\0\0\202\2\7\320\343\6H\26P\340\277\370\330\30e\30"..., 4096) = 4096
26882 10:38:41.920229 lseek(3, 848175104, SEEK_SET) = 848175104
26882 10:38:41.920245 write(3, "\220\240@\6\371\341\277>\0\2402\24\0\0\0\0\202\2\7\320\343\6H\26P\340\277\370\330\30e\30"..., 4096) = 4096
26882 10:38:41.920296 lseek(3, 1060147200, SEEK_SET) = 1060147200
26882 10:38:41.920312 write(3, "\220\240@\6\371\341\277>\0\2402\24\0\0\0\0\202\2\7\320\343\6H\26P\340\277\370\330\30e\30"..., 4096) = 4096
26882 10:38:41.920362 lseek(3, 863690752, SEEK_SET) = 863690752
26882 10:38:41.920377 write(3, "\220\240@\6\371\341\277>\0\2402\24\0\0\0\0\202\2\7\320\343\6H\26P\340\277\370\330\30e\30"..., 4096) = 4096
26882 10:38:41.920428 lseek(3, 279457792, SEEK_SET) = 279457792
26882 10:38:41.920444 write(3, "\220\240@\6\371\341\277>\0\2402\24\0\0\0\0\202\2\7\320\343\6H\26P\340\277\370\330\30e\30"..., 4096) = 4096
26882 10:38:41.920492 lseek(3, 271794176, SEEK_SET) = 271794176
26882 10:38:41.920508 write(3, "\220\240@\6\371\341\277>\0\2402\24\0\0\0\0\202\2\7\320\343\6H\26P\340\277\370\330\30e\30"..., 4096) = 4096
26882 10:38:41.920558 lseek(3, 1067864064, SEEK_SET) = 1067864064
26882 10:38:41.920573 write(3, "\220\240@\6\371\341\277>\0\2402\24\0\0\0\0\202\2\7\320\343\6H\26P\340\277\370\330\30e\30"..., 4096) = 4096
...The current file offset is located through lseek before every write of random read
Synchronous sequential writing
Test sequential writing in the way just now
root@wilson-ubuntu:~# strace -f -tt -o /tmp/write.log -D fio -name=write -rw=write \
-direct=1 -bs=4k -size=1G -numjobs=1 -group_reporting -filename=/tmp/test.db
write: (g=0): rw=write, bs=4K-4K/4K-4K/4K-4K, ioengine=sync, iodepth=1
fio-2.2.10
Starting 1 process
Jobs: 1 (f=1): [W(1)] [100.0% done] [0KB/70432KB/0KB /s] [0/17.7K/0 iops] [eta 00m:00s]
write: (groupid=0, jobs=1): err= 0: pid=27005: Wed Aug 14 10:53:02 2019
write: io=1024.0MB, bw=70238KB/s, iops=17559, runt= 14929msec
clat (usec): min=43, max=7464, avg=55.95, stdev=56.24
lat (usec): min=43, max=7465, avg=56.15, stdev=56.25
...
bw (KB /s): min=67304, max=72008, per=99.98%, avg=70225.38, stdev=1266.88
...
Run status group 0 (all jobs):
WRITE: io=1024.0MB, aggrb=70237KB/s, minb=70237KB/s, maxb=70237KB/s, mint=14929msec, maxt=14929msec
Disk stats (read/write):
...
sda: ios=0/262162, merge=0/10, ticks=0/6948, in_queue=6932, util=46.49%
You can see:
Throughput increased to about 70M
Let's look at 1 kernel call again:
root@wilson-ubuntu:~# more /tmp/write.log
...
27046 10:54:28.194508 write(3, "\0\0\23\0\0\0\0\0\0\300\16\0\0\0\0\0\0\360\t\0\0\0\0\0\0\320\17\0\0\0\0\0"..., 4096) = 4096
27046 10:54:28.194568 write(3, "\0\0\23\0\0\0\0\0\0\300\16\0\0\0\0\0\0\220\24\0\0\0\0\0\0\320\17\0\0\0\0\0"..., 4096) = 4096
27046 10:54:28.194627 write(3, "\0\0\23\0\0\0\0\0\0\300\16\0\0\0\0\0\0\220\24\0\0\0\0\0\0\320\17\0\0\0\0\0"..., 4096) = 4096
27046 10:54:28.194687 write(3, "\0\0\23\0\0\0\0\0\0\300\16\0\0\0\0\0\0\220\24\0\0\0\0\0\0\320\17\0\0\0\0\0"..., 4096) = 4096
27046 10:54:28.194747 write(3, "\0\0\23\0\0\0\0\0\0\300\16\0\0\0\0\0\0\220\24\0\0\0\0\0\0\320\17\0\0\0\0\0"..., 4096) = 4096
27046 10:54:28.194807 write(3, "\0\0\23\0\0\0\0\0\0\300\16\0\0\0\0\0\0\220\24\0\0\0\0\0\0\320\17\0\0\0\0\0"..., 4096) = 4096
27046 10:54:28.194868 write(3, "\0\0\23\0\0\0\0\0\0\300\16\0\0\0\0\0\0\220\24\0\0\0\0\0\0\320\17\0\0\0\0\0"..., 4096) = 4096
27046 10:54:28.194928 write(3, "\0\0\23\0\0\0\0\0\0\300\16\0\0\0\0\0\0\220\24\0\0\0\0\0\0\320\17\0\0\0\0\0"..., 4096) = 4096
27046 10:54:28.194988 write(3, "\0\0\23\0\0\0\0\0\0\300\16\0\0\0\0\0\0\220\24\0\0\0\0\0\0\320\17\0\0\0\0\0"..., 4096) = 4096
27046 10:54:28.195049 write(3, "\0\0\23\0\0\0\0\0\0\300\16\0\0\0\0\0\0\220\24\0\0\0\0\0\0\320\17\0\0\0\0\0"..., 4096) = 4096
27046 10:54:28.195110 write(3, "\0\0\23\0\0\0\0\0\0\300\16\0\0\0\0\0\0\220\24\0\0\0\0\0\0\320\17\0\0\0\0\0"..., 4096) = 4096
27046 10:54:28.195197 write(3, "\0\0\23\0\0\0\0\0\0\300\16\0\0\0\0\0\0\220\24\0\0\0\0\0\0\320\17\0\0\0\0\0"..., 4096) = 4096
27046 10:54:28.195262 write(3, "\0\0\23\0\0\0\0\0\0\300\16\0\0\0\0\0\0\220\24\0\0\0\0\0\0\320\17\0\0\0\0\0"..., 4096) = 4096
27046 10:54:28.195330 write(3, "\0\0\23\0\0\0\0\0\0\300\16\0\0\0\0\0\0\220\24\0\0\0\0\0\0\320\17\0\0\0\0\0"..., 4096) = 4096
27046 10:54:28.195426 write(3, "\0\0\23\0\0\0\0\0\0\300\16\0\0\0\0\0\0\220\24\0\0\0\0\0\0\320\17\0\0\0\0\0"..., 4096) = 4096
27046 10:54:28.195497 write(3, "\0\0\23\0\0\0\0\0\0\300\16\0\0\0\0\0\0\220\24\0\0\0\0\0\0\320\17\0\0\0\0\0"..., 4096) = 4096
27046 10:54:28.195567 write(3, "\0\0\23\0\0\0\0\0\0\300\16\0\0\0\0\0\0\220\24\0\0\0\0\0\0\320\17\0\0\0\0\0"..., 4096) = 4096
27046 10:54:28.195637 write(3, "\0\0\23\0\0\0\0\0\0\300\16\0\0\0\0\0\0\220\24\0\0\0\0\0\0\320\17\0\0\0\0\0"..., 4096) = 4096
27046 10:54:28.195704 write(3, "\0\0\23\0\0\0\0\0\0\300\16\0\0\0\0\0\0\220\24\0\0\0\0\0\0\320\17\0\0\0\0\0"..., 4096) = 4096
27046 10:54:28.195757 write(3, "\0\0\23\0\0\0\0\0\0\300\16\0\0\0\0\0\0\220\24\0\0\0\0\0\0\320\17\0\0\0\0\0"..., 4096) = 4096
27046 10:54:28.195807 write(3, "\0\0\23\0\0\0\0\0\0\300\16\0\0\0\0\0\0\220\24\0\0\0\0\0\0\320\17\0\0\0\0\0"..., 4096) = 4096
27046 10:54:28.195859 write(3, "\0\0\23\0\0\0\0\0\0\300\16\0\0\0\0\0\0\220\24\0\0\0\0\0\0\320\17\0\0\0\0\0"..., 4096) = 4096
27046 10:54:28.195910 write(3, "\0\0\23\0\0\0\0\0\0\300\16\0\0\0\0\0\0\220\24\0\0\0\0\0\0\320\17\0\0\0\0\0"..., 4096) = 4096
27046 10:54:28.195961 write(3, "\0\0\23\0\0\0\0\0\0\300\16\0\0\0\0\0\0\220\24\0\0\0\0\0\0\320\17\0\0\0\0\0"..., 4096) = 4096
27046 10:54:28.196012 write(3, "\0\0\23\0\0\0\0\0\0\300\16\0\0\0\0\0\0\220\24\0\0\0\0\0\0\320\17\0\0\0\0\0"..., 4096) = 4096
27046 10:54:28.196062 write(3, "\0\0\23\0\0\0\0\0\0\300\16\0\0\0\0\0\0\220\24\0\0\0\0\0\0\320\17\0\0\0\0\0"..., 4096) = 4096
27046 10:54:28.196112 write(3, "\0\0\23\0\0\0\0\0\0\300\16\0\0\0\0\0\0 \26\0\0\0\0\0\0\320\17\0\0\0\0\0"..., 4096) = 4096
27046 10:54:28.196162 write(3, "\0\0\23\0\0\0\0\0\0\300\16\0\0\0\0\0\0 \26\0\0\0\0\0\0\320\17\0\0\0\0\0"..., 4096) = 4096
27046 10:54:28.196213 write(3, "\0\0\23\0\0\0\0\0\0\300\16\0\0\0\0\0\0 \26\0\0\0\0\0\0\320\17\0\0\0\0\0"..., 4096) = 4096
27046 10:54:28.196265 write(3, "\0\0\23\0\0\0\0\0\0\300\16\0\0\0\0\0\0 \26\0\0\0\0\0\0\320\17\0\0\0\0\0"..., 4096) = 4096
27046 10:54:28.196314 write(3, "\0\0\23\0\0\0\0\0\0\300\16\0\0\0\0\0\0 \26\0\0\0\0\0\0\320\17\0\0\0\0\0"..., 4096) = 4096
27046 10:54:28.196363 write(3, "\0\0\23\0\0\0\0\0\0\300\16\0\0\0\0\0\0 \26\0\0\0\0\0\0\320\17\0\0\0\0\0"..., 4096) = 4096
27046 10:54:28.196414 write(3, "\0\0\23\0\0\0\0\0\0\300\16\0\0\0\0\0\0 \26\0\0\0\0\0\0\320\17\0\0\0\0\0"..., 4096) = 4096
27046 10:54:28.196472 write(3, "\0\0\23\0\0\0\0\0\0\300\16\0\0\0\0\0\0 \26\0\0\0\0\0\0\320\17\0\0\0\0\0"..., 4096) = 4096
27046 10:54:28.196524 write(3, "\0\0\23\0\0\0\0\0\0\300\16\0\0\0\0\0\0 \26\0\0\0\0\0\0\320\17\0\0\0\0\0"..., 4096) = 4096
27046 10:54:28.196573 write(3, "\0\0\23\0\0\0\0\0\0\300\16\0\0\0\0\0\0 \26\0\0\0\0\0\0\320\17\0\0\0\0\0"..., 4096) = 4096
...Because of sequential reading, there is no need to repeatedly locate file offsets, so you can focus on writing operations
5. slat indicators
From the above test, we did not find slat in the test report of fio, because the above are synchronous operations. For synchronous I/O, since I/O submission and I/O completion are one action, slat is actually the completion time of I/O
Asynchronous sequential write, add the command for synchronous sequential write-ioengine=libaio:
root@wilson-ubuntu:~# fio -name=write -rw=write -ioengine=libaio -direct=1 -bs=4k -size=1G -numjobs=1 -group_reporting -filename=/tmp/test.db
write: (g=0): rw=write, bs=4K-4K/4K-4K/4K-4K, ioengine=libaio, iodepth=1
fio-2.2.10
Starting 1 process
Jobs: 1 (f=1): [W(1)] [100.0% done] [0KB/119.3MB/0KB /s] [0/30.6K/0 iops] [eta 00m:00s]
write: (groupid=0, jobs=1): err= 0: pid=27258: Wed Aug 14 11:14:36 2019
write: io=1024.0MB, bw=120443KB/s, iops=30110, runt= 8706msec
slat (usec): min=3, max=70, avg= 4.31, stdev= 1.56
clat (usec): min=0, max=8967, avg=28.13, stdev=55.68
lat (usec): min=22, max=8976, avg=32.53, stdev=55.72
...
bw (KB /s): min=118480, max=122880, per=100.00%, avg=120467.29, stdev=1525.68
...
Run status group 0 (all jobs):
WRITE: io=1024.0MB, aggrb=120442KB/s, minb=120442KB/s, maxb=120442KB/s, mint=8706msec, maxt=8706msec
Disk stats (read/write):
...
sda: ios=0/262147, merge=0/1, ticks=0/6576, in_queue=6568, util=74.32%It can be seen that slat index appears, and lat is approximately equal to the sum of slat + clat (avg average). After the asynchronous io is replaced, the throughput has been greatly improved, about 120M
6. Summary
fio should be used as the baseline tool of disk. When you get a machine (physical machine or cloud machine), you should do a baseline test on the disk of the machine in the first time, so as to know fairly well
All the tests in this paper bypass the cache, and the influence of cache should be taken into account in practical application